def __init__(self, topN=4, classNum=200):
super(attention_net, self).__init__()
# self.pretrained_model = resnet.resnet50(pretrained=True)
# self.pretrained_model = resnet.resnet152(pretrained=True)
# self.pretrained_model = resnet.resnet101(pretrained=True)
# self.pretrained_model = densenet.densenet121(pretrained=False,num_classes=classNum)
# self.pretrained_model = densenet.densenet169(pretrained=True,num_classes=classNum)
# self.pretrained_model = densenet.densenet161(pretrained=True,num_classes=classNum)
# self.pretrained_model = resnext.resnext101_32x8d(pretrained=True,num_classes=classNum)
# self.pretrained_model = mobilenet.mobilenet_v2(pretrained=True,num_classes=classNum)
self.pretrained_model = squeezenet.squeezenet1_1(pretrained=True,
num_classes=classNum)
# self.pretrained_model.avgpool = nn.AdaptiveAvgPool2d(1)
# # print("classNum: ",classNum) # classNum: 20 Fish
# num_ftrs = self.pretrained_model.fc.in_features
# self.pretrained_model.fc = nn.Linear(num_ftrs, classNum)
# self.pretrained_model.fc = nn.Linear(512 * 4, classNum)
self.proposal_net = ProposalNet(in_features=512)
self.topN = topN
self.concat_net = nn.Linear(5000, classNum) # 5000
self.partcls_net = nn.Linear(1, classNum) # 1
_, edge_anchors, _ = generate_default_anchor_maps()
self.pad_side = 224
# self.edge_anchors = (edge_anchors + 224).astype(np.int)
self.edge_anchors = (edge_anchors + 224).astype(np.int).repeat(
4, 0)[:5991, :]
def __init__(self, topN=4):
super(attention_net, self).__init__()
# ResNet family
#self.pretrained_model = resnet.resnet50(pretrained=True)
#self.pretrained_model = resnet.resnet101(pretrained=True)
#self.pretrained_model = resnet.resnet152(pretrained=True)
# ResNext
self.pretrained_model = resnext.resnext50_32x4d(pretrained=True)
#self.pretrained_model = resnext.resnext101_32x8d(pretrained=True)
self.pretrained_model.avgpool = nn.AdaptiveAvgPool2d(1)
self.pretrained_model.fc = nn.Linear(512 * 4, CLASSES)
# initiate FPN, Scrutinizer, Teacher net
self.proposal_net = ProposalNet()
self.topN = topN
self.concat_net = nn.Linear(2048 * (CAT_NUM + 1), CLASSES)
self.partcls_net = nn.Linear(512 * 4, CLASSES)
# use edge_anchors
_, edge_anchors, _ = generate_default_anchor_maps()
self.pad_side = PAD_SIZE
self.edge_anchors = (edge_anchors + PART_IMAGE_SIZE).astype(np.int)
def __init__(self, topN=4, classNum=200):
super(attention_net, self).__init__()
# self.pretrained_model = resnet.resnet50(pretrained=True)
# self.pretrained_model = resnext.resnet18(pretrained=True,num_classes=classNum)
# self.pretrained_model = resnext.resnet34(pretrained=True,num_classes=classNum)
# self.pretrained_model = resnet.resnet152(pretrained=True)
# self.pretrained_model = resnet.resnet101(pretrained=True)
# self.pretrained_model = densenet.densenet121(pretrained=False,num_classes=classNum)
# self.pretrained_model = densenet.densenet169(pretrained=True,num_classes=classNum)
# self.pretrained_model = densenet.densenet161(pretrained=True,num_classes=classNum)
# self.pretrained_model = resnext.resnext101_32x8d(pretrained=True,num_classes=classNum)
# self.pretrained_model = resnext.wide_resnet101_2(pretrained=True,num_classes=classNum)
# self.pretrained_model.avgpool = nn.AdaptiveAvgPool2d(1)
# print("classNum: ",classNum) # classNum: 20 Fish
self.pretrained_model = vgg.vgg16(pretrained=True,
num_classes=classNum)
# num_ftrs = self.pretrained_model.classifier.in_features
# self.pretrained_model.fc = nn.Linear(num_ftrs, classNum)
# self.pretrained_model.fc = nn.Linear(512 * 4, classNum)
self.proposal_net = ProposalNet()
self.topN = topN
self.concat_net = nn.Linear(125440, classNum)
self.partcls_net = nn.Linear(25088, classNum)
_, edge_anchors, _ = generate_default_anchor_maps()
self.pad_side = 224
self.edge_anchors = (edge_anchors + 224).astype(np.int)
def __init__(self, topN=4, n_class=200):
super(attention_net, self).__init__()
self.n_class = n_class
self.pretrained_model = resnet.resnet50(pretrained=True)
self.pretrained_model.avgpool = nn.AdaptiveAvgPool2d(1)
self.pretrained_model.fc = nn.Linear(512 * 4, self.n_class)
self.proposal_net = ProposalNet()
self.topN = topN
self.concat_net = nn.Linear(2048 * (CAT_NUM + 1 + 1), self.n_class)
self.partcls_net = nn.Linear(512 * 4, self.n_class)
self.pad_side = 224
_, edge_anchors_small, _ = generate_default_anchor_maps(
setting='small')
self.edge_anchors_small = (edge_anchors_small + 224).astype(np.int)
_, edge_anchors_large, _ = generate_default_anchor_maps(
setting='large')
self.edge_anchors_large = (edge_anchors_large + 224).astype(np.int)
def __init__(self, topN=4):
super(attention_net, self).__init__()
self.pretrained_model = resnet.resnet50(pretrained=True)
self.pretrained_model.avgpool = nn.AdaptiveAvgPool2d(1)
self.pretrained_model.fc = nn.Linear(512 * 4, 200)
self.proposal_net = ProposalNet()
self.topN = topN
self.concat_net = nn.Linear(2048 * (CAT_NUM + 1), 200)
self.partcls_net = nn.Linear(512 * 4, 200)
_, edge_anchors, _ = generate_default_anchor_maps()
self.pad_side = 224
self.edge_anchors = (edge_anchors + 224).astype(np.int)
def __init__(self, topN=4):
super(attention_net, self).__init__()
self.pretrained_model = resnet.resnet50(pretrained=True)
self.pretrained_model.add_module('raw_fc',
nn.Linear(512 * 4, LABEL_NUM))
self.proposal_net = ProposalNet()
self.topN = topN
self.concat_net = nn.Linear(2048 * (CAT_NUM + 1), LABEL_NUM)
self.partcls_net = nn.Linear(512 * 4, LABEL_NUM)
_, edge_anchors, _ = generate_default_anchor_maps()
self.pad_side = 224
self.edge_anchors = (edge_anchors + 224).astype(np.int)
def __init__(self, topN=4):
super(attention_net, self).__init__()
self.pretrained_model = resnet.resnet152(pretrained=True)
#self.pretrained_model = pretrainedmodels.__dict__['se_resnext101_32x4d'](num_classes=1000, pretrained=None)
self.pretrained_model.avgpool = nn.AdaptiveAvgPool2d(1)
self.pretrained_model.last_linear = nn.Linear(512 * 4, 2)
self.proposal_net = ProposalNet()
self.topN = topN
self.concat_net = nn.Linear(2048 * (CAT_NUM + 1), 2)
self.partcls_net = nn.Linear(512 * 4, 2)
_, edge_anchors, _ = generate_default_anchor_maps()
self.pad_side = 224
self.edge_anchors = (edge_anchors + 224).astype(np.int)
def __init__(self, topN=4):
super(attention_net, self).__init__()
self.pretrained_model = densenet.densenet201(pretrained=True)
self.in_features = self.pretrained_model.classifier.in_features
# self.pretrained_model.avgpool = nn.AdaptiveAvgPool2d(1)
self.pretrained_model.classifier = nn.Linear(1920, 200)
self.proposal_net = ProposalNet()
self.topN = topN
self.concat_net = nn.Linear(self.in_features* (CAT_NUM + 1), 200)
self.partcls_net = nn.Linear(1920, 200)
_, edge_anchors, _ = generate_default_anchor_maps()
self.pad_side = 224
self.edge_anchors = (edge_anchors + 224).astype(np.int)
def __init__(self, topN=4):
super(attention_net, self).__init__()
self.pretrained_model = resnet.resnet50(pretrained=True)
self.pretrained_model.avgpool = nn.AdaptiveAvgPool2d(1)
self.pretrained_model.fc = nn.Linear(512 * 4, NUM_CLASS)
self.proposal_net = ProposalNet()
self.topN = topN # proposal number
self.concat_net = nn.Linear(2048 * (CAT_NUM + 1), NUM_CLASS)
self.partcls_net = nn.Linear(512 * 4, NUM_CLASS)
#其中每行的4个值[公式]表示矩形左上角和右下角点坐标。9个矩形共有3种形状，3种长宽比
_, edge_anchors, _ = generate_default_anchor_maps()
self.pad_side = 224
self.edge_anchors = (edge_anchors + 224).astype(np.int)
def __init__(self, topN=4, classNum=20):
super(attention_net, self).__init__()
# self.pretrained_model = shufflenetv2.shufflenet_v2_x0_5(pretrained=True,num_classes=classNum)
self.pretrained_model = shufflenetv2.shufflenet_v2_x1_0(
pretrained=True, num_classes=classNum)
self.proposal_net = ProposalNet(in_features=24)
self.topN = topN
self.concat_net = nn.Linear(5120, classNum)
self.partcls_net = nn.Linear(1024, classNum)
_, edge_anchors, _ = generate_default_anchor_maps()
self.pad_side = 224
self.edge_anchors = (edge_anchors + 224).astype(np.int).repeat(
251, 0)[:404544, :]
def __init__(self, topN=4):
super(attention_net, self).__init__()
# self.pretrained_model = densenet.densenet201(pretrained=True)
self.pretrained_model = resnet.resnet50(pretrained=True)
self.pretrained_model.avgpool = nn.AdaptiveAvgPool2d(
1) # 1 denotes the output size
num_features = self.pretrained_model.num_features
self.pretrained_model.fc = nn.Linear(
num_features, NUM_CLS) # in_features=512*4, out_features=200
self.proposal_net = ProposalNet(num_features)
self.topN = topN
self.concat_net = nn.Linear(num_features * (CAT_NUM + 1), NUM_CLS)
self.partcls_net = nn.Linear(num_features, NUM_CLS)
_, edge_anchors, _ = generate_default_anchor_maps()
self.pad_side = 224
self.edge_anchors = (edge_anchors + 224).astype(np.int)
def __init__(self, topN=4, classNum=20):
super(attention_net, self).__init__()
# self.pretrained_model = resnet.resnet50(pretrained=True)
# self.pretrained_model = resnet.resnet152(pretrained=True)
self.pretrained_model = inception.inception_v3(pretrained=True)
# self.pretrained_model.avgpool = nn.AdaptiveAvgPool2d(1)
# self.pretrained_model.fc = nn.Linear(512 * 4, 200)
num_ftrs = self.pretrained_model.fc.in_features
self.pretrained_model.fc = nn.Linear(num_ftrs, classNum)
self.proposal_net = ProposalNet()
self.topN = topN
self.concat_net = nn.Linear(2048 * (CAT_NUM + 1), classNum)
self.partcls_net = nn.Linear(512 * 4, classNum)
_, edge_anchors, _ = generate_default_anchor_maps()
self.pad_side = 224
self.edge_anchors = (edge_anchors + 224).astype(np.int)
def __init__(self, topN=4, classNum=20):
super(attention_net, self).__init__()
# self.pretrained_model = resnet.resnet50(pretrained=True)
# self.pretrained_model = resnet.resnet152(pretrained=True)
# self.pretrained_model = resnet.resnet101(pretrained=True)
self.pretrained_model = densenet.densenet121(pretrained=False,
num_classes=classNum)
# self.pretrained_model.avgpool = nn.AdaptiveAvgPool2d(1)
# print("classNum: ",classNum) # classNum: 20 Fish
num_ftrs = self.pretrained_model.classifier.in_features
self.pretrained_model.classifier = nn.Linear(num_ftrs, classNum)
# self.pretrained_model.fc = nn.Linear(512 * 4, classNum)
self.proposal_net = ProposalNet()
self.topN = topN
self.concat_net = nn.Linear(1024 * (CAT_NUM + 1), classNum)
self.partcls_net = nn.Linear(512 * 2, classNum)
_, edge_anchors, _ = generate_default_anchor_maps()
self.pad_side = 224
self.edge_anchors = (edge_anchors + 224).astype(np.int)
